Learning from Multiple Noisy Partial Labelers

被引:0
|
作者
Yu, Peilin [1 ]
Ding, Tiffany [2 ]
Bach, Stephen H. [1 ]
机构
[1] Brown Univ, Providence, RI 02912 USA
[2] Univ Calif Berkeley, Berkeley, CA USA
关键词
D O I
暂无
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
Programmatic weak supervision creates models without hand-labeled training data by combining the outputs of heuristic labelers. Existing frameworks make the restrictive assumption that labelers output a single class label. Enabling users to create partial labelers that output subsets of possible class labels would greatly expand the expressivity of programmatic weak supervision. We introduce this capability by defining a probabilistic generative model that can estimate the underlying accuracies of multiple noisy partial labelers without ground truth labels. We show how to scale up learning, for example learning on 100k examples in one minute, a 300x speed up compared to a naive implementation. We also prove that this class of models is generically identifiable up to label swapping under mild conditions. We evaluate our framework on three text classification and six object classification tasks. On text tasks, adding partial labels increases average accuracy by 8.6 percentage points. On image tasks, we show that partial labels allow us to approach some zero-shot object classification problems with programmatic weak supervision by using class attributes as partial labelers. On these tasks, our framework has accuracy comparable to recent embedding-based zero-shot learning methods, while using only pre-trained attribute detectors.
引用
收藏
页数:24
相关论文
共 50 条
  • [41] Learning to Rank from Noisy Data
    Ding, Wenkui
    Geng, Xiubo
    Zhang, Xu-Dong
    ACM TRANSACTIONS ON INTELLIGENT SYSTEMS AND TECHNOLOGY, 2015, 7 (01)
  • [42] MACHINE LEARNING FROM NOISY INFORMATION
    FINDLER, NV
    NATURE, 1964, 204 (495) : 103 - &
  • [43] Learning from Noisy Label Distributions
    Yoshikawa, Yuya
    ARTIFICIAL NEURAL NETWORKS AND MACHINE LEARNING, PT II, 2017, 10614 : 137 - 145
  • [44] Learning Programs from Noisy Data
    Raychev, Veselin
    Bielik, Pavol
    Vechev, Martin
    Krause, Andreas
    ACM SIGPLAN NOTICES, 2016, 51 (01) : 761 - 774
  • [45] Learning programs from noisy data
    Raychev V.
    Bielik P.
    Vechev M.
    Krause A.
    1600, Association for Computing Machinery, 2 Penn Plaza, Suite 701, New York, NY 10121-0701, United States (51): : 761 - 774
  • [46] Learning from Noisy Labels with Distillation
    Li, Yuncheng
    Yang, Jianchao
    Song, Yale
    Cao, Liangliang
    Luo, Jiebo
    Li, Li-Jia
    2017 IEEE INTERNATIONAL CONFERENCE ON COMPUTER VISION (ICCV), 2017, : 1928 - 1936
  • [47] Learning with Noisy Partial Labels by Simultaneously Leveraging Global and Local Consistencies
    Li, Changchun
    Li, Ximing
    Ouyang, Jihong
    CIKM '20: PROCEEDINGS OF THE 29TH ACM INTERNATIONAL CONFERENCE ON INFORMATION & KNOWLEDGE MANAGEMENT, 2020, : 725 - 734
  • [48] Learning partial differential equations for biological transport models from noisy spatio-temporal data
    Lagergren, John H.
    Nardini, John T.
    Michael Lavigne, G.
    Rutter, Erica M.
    Flores, Kevin B.
    PROCEEDINGS OF THE ROYAL SOCIETY A-MATHEMATICAL PHYSICAL AND ENGINEERING SCIENCES, 2020, 476 (2234):
  • [49] Learning from Noisy Data with Robust Representation Learning
    Li, Junnan
    Xiong, Caiming
    Hoi, Steven C. H.
    2021 IEEE/CVF INTERNATIONAL CONFERENCE ON COMPUTER VISION (ICCV 2021), 2021, : 9465 - 9474
  • [50] Learning from Multiple Annotator Noisy Labels via Sample-Wise Label Fusion
    Gao, Zhengqi
    Sun, Fan-Keng
    Yang, Mingran
    Ren, Sucheng
    Xiong, Zikai
    Engeler, Marc
    Burazer, Antonio
    Wildling, Linda
    Daniel, Luca
    Boning, Duane S.
    COMPUTER VISION, ECCV 2022, PT XXIV, 2022, 13684 : 407 - 422